In the spring of 2000 I had the opportunity to write a two-part article for Gamasutra.com, regarding game companies benefiting from the lessons learned in the theme park industry for the designs of their 3D virtual worlds. The articles had a favorable response from a large community of game designers, who had been formulating similar philosophies for their own 3D projects.

Another result of the article was a call I received from a then startup company in Menlo Park, CA, called There, Inc. It was intending to tackle a project that would allow me to practice the very design philosophies I was preaching. This began a four year odyssey that would allow me to test every design principle I had outlined, as we endeavored to plan everything from continent-sized landmasses to virtual T-shirt graphics.

After four years of art directing the There.com art team, I emerged having learned many new lessons about environment design. Although the principles listed in my previous articles held true, the unique character of a massively multiplayer virtual world startup created surprising challenges, ones that would stretch our team's ability to create a truly immersive online environment. Below, I have listed just a few of these lessons for your review and edification:

THE LESSONS

Translating Concrete into Polygons

While working on theme park projects, I constantly bumped my head on the inevitable ceiling of our project's budget, and the limits of the physical world. Walt Disney once complained that all of the money that had been raised for the construction of Disneyland was being spent on necessities that would never be seen by the public. The infrastructure of any construction project, namely the plumbing, wiring, sewage, and other facility necessities, although invisible, is the price you must pay if you intend to make your dream a physical reality. As a project gets closer and closer to completion, a designer's ability to change or add anything is constrained by the decisions made much earlier in the production cycle.

While working on the construction site of any project, you frequently come upon details that, once seen full scale and within the context of the physical space, might have been better if only a few slight alterations could have been agreed upon. During the early design phase of your project, changes of this sort are a necessary part of the evolution of any design.

In construction, however, even small changes become monumental tasks, which constantly threaten to impact the budget and schedule of your project. Moving a light fixture from one wall to another is no problem, when it is in pencil or within a CAD document. In the real world, moving that same fixture can mean hours of lost time as new conduit is pulled, holes are drilled, circuits rerouted, and previous holes patched. The reality of most construction sites is coming to terms with past decisions you have made, and learning to love them (or at least pretending that you do).

Another facet to the reality of watching your designs born into concrete is to realize that what works on paper doesn't necessarily always translate to plywood and lathe. As various vendors arrive with their many bits and pieces, you are confronted with the unpleasant truth that not everything "fits" as you had designed. Doors are slightly too big, window proportions are not as elegant as you had imagined, and shortcuts you had made in the number of stair steps leading to an elevated area are becoming a dangerous trip hazard. As each problem arises you are put on the spot to rethink your design on the fly, or compromise your initial vision for the sake of usability. As these compromises mount, and as you get ever closer to your project's deadline, you find yourself waxing dreamily on the potential of a non-physical environment to design within. The mind reels at the prospect of moving polygons rather then concrete in your virtual construction site, to be able to relocate digital light fixtures with a simple cut & paste, without costly overtime from a union electrician.

Under these circumstances the possibilities of designing environments for virtual worlds seems to be the perfect answer for a designer frustrated by the limits of expensive building materials and the reality of dealing with things like, well… gravity.

Awakening from the Dream

I entered my new position as the designer of a digital world with all the glee of a child promised that Christmas would be coming every day. No longer would I be shackled by the frustrating realities of design limited by physical constraints. Finally I would be able to make design decisions up to and beyond the eleventh hour. Virtual world design was going to be sweet, where your building materials are always pliable and where whim rather than unions dictate alterations.

This dream was unfortunately short-lived. It took several weeks, but I eventually came to understand that although the limits were different, the limits did exist. While on the physical construction site, a designer is limited to the constraints of finances and time, the virtual designer's limits are just as tangible.

While gravity is no longer a problem worth consideration, you are, however the slave of another constraint… the limits of your target CPU. Even though computers seem to be growing more powerful every other week, the reality is that computers can only crunch so many numbers at a time. With the addition of a 3D accelerator card installed in your computer, you boost that ability quite a bit, but the fact remains that your rich 3D world can only have the amount of geometry in it that your computer can "think" about at any given time.

In the simplest terms, this means that anything you "see" from within your 3D world is generated because your computer is doing the math necessary to display it for you. This frequently includes not only what is visible, but also what might exist just out of sight, behind you, or just over the hill from where you are standing. Radiating from your standing position, in very direction, your CPU is crunching away an insane amount of information in preparation that the player's free will might cause you to move or look in any direction. Like those invisible sewage pipes and the electrical conduit, your CPU budget is being drained by buildings, trees, and people you can't even see from where you are standing. Not only that, but if you are building an online environment, your budget is also constrained by anyone who MIGHT show up. The punishment for not paying close attention to these budgetary limits is that once exceeded, your audience's experience begins to suffer. Your CPU can only "see" so many polygons at a time, and if there is more geometry then it can compute, then your framerate begins to drop, things start to disappear, and worst of all, other people cease to exist.

The Price of Everything Else

In the physical world, we don't think twice about the existence of things like sunlight, water, or wildlife, as these are purely the perks of the real world. In the digital world however, the mere existence of these elements means that they need to have been fabricated and rendered by your computer. Like the mounting cost of geometry on your CPU brainpower, light, weather effects, and the AI of animated fauna can cause a huge hit on your computer's limited budget. Even an empty room constructed with a meager number of polygons can cause your framerate to plummet if it includes only a few animated light fixtures. Gone is the dream of Cut & Paste light fixtures, when the addition of such an effect might necessitate you moving every bit of furniture out of the room to accommodate it.

Learning to "Love" Your Limits

My first encounter with designing for a multiplayer online world was the challenge of learning to come to terms with these limitations, and finding clever ways to design in spite of them. Most single-player, or even limited multi-player games, needs to accommodate the possibility of a dozen or so additional characters or avatars that might appear in any given environment. This leaves a budget that allows their designers the luxury to build detail-rich spaces, which takes advantage of the effects today's powerful 3D cards can deliver. Furthermore, most game companies can depend on an audience that is routinely willing to upgrade their systems to meet the demands of cutting-edge titles. In our case, our product needed to potentially accommodate the arrival of 50 to 100 avatars in an environment, and was marketed to a demographic of computer owners that have never even heard of a 3D card. With a budget of 1500 or more polygons per avatar, and the potential of hundreds congregating in any given space, this left the environmental designers a budget of no more then 10,000 polys per any given virtual location.

Reeling under these limitations, we worked to become the living definition of "less is more". Equally hard hit was our texture budget, which insisted that since our member avatars could show up wearing countless numbers of diverse clothing textures, our building textures would need to be just as minimal as our structures. Under these constraints, we developed a graphic style which chose visual consistency over complexity, and immersion through suggested context rather then spelling out every detail. Although our efforts were applauded, this choice made visiting each annual E3 Conference a painful pilgrimage of what is possible for every 3D game but ours.

One comforting realization came when exploring the 3D worlds of other companies attempting similar online multi-player environments. Whether you are visiting the streets of Toontown.com or the planets of Star Wars Galaxies, you begin to see that each game has come up with solutions based on similar limits. Whether they chose to allow their trees to render only when you are right on top of them, or they limited vertical movement because their props are flat textured "billboards", each design team did their very best to immerse their audiences despite how little they have to build with.